Thursday, June 22, 2006

A version of net neutrality I can endorse

In an attempt to offer something constructive, here's a version of network neutrality--let's call it Lippard Network Neutrality--that seems to me to be reasonable, providing me with what I want as a consumer of Internet services and what I would want if I were managing security for the provider of those services:

1. Nondiscrimination

Companies that provide facilities-based wireline broadband (i.e., those who own the last-mile wires) to residences must provide unrestricted Internet access to their customers who wish to purchase Internet access, allowing the use of any Internet service or application that does not violate any laws or cause degradation or disruption to the service or other customers. The provider may engage in filtering for consumer-grade service in order to prevent the spread of malware and the sending of spam, including (for example) SMTP filtering or redirection to the provider's mail services, but must allow the purchase of business-grade service under which customers may operate their own mail servers. The provider retains the right to suspend service or quarantine users that send spam, become compromised with malware, or engage in illegal activity or activity that disrupts the service.

2. Unbundling

Providers must unbundle Internet access from other services sold over the same connection, so that a customer may use the entire capacity of the circuit for Internet access.

These two requirements would give me what I want as a customer, as well as give the provider the ability to recover their costs, provide services that use QoS, provide additional filtering to protect their network and the rest of their customer base from malware, and so on. I think it's quite reasonable for a basic consumer Internet service to do port 25 filtering, force the use of the provider's mail servers, and to do network-based filtering of malware--but I would like the ability to pay extra for completely unfiltered Internet service and take steps to protect myself. And in fact, that's what I'm currently paying Cox for today--I pay for business-grade service to my home in order to run my own servers here, though I could put those servers into a colo facility and get the same effect, which is what I would do if Cox decided to discontinue offering business-class service to residences. Because that option exists, it would not be necessary to mandate that providers must provide business class service as I described above, but I'd still want to be able to ensure that I could access my remotely hosted services from home.

How this differs from what many network neutrality advocates are arguing for:

1. I don't prohibit QoS or tiering, as that is a genuinely useful network feature where I expect to see future innovation of services that depend on it.
2. The nondiscrimination provision is written to allow some kind of less-than-full-Internet walled garden service at low cost--so long as customers can still purchase real Internet service. (I think such a service would be under competitive pressure to allow access to the full Internet, for the same reason AOL ended up allowing full Internet access--otherwise the service wouldn't attract enough users to be a successful product offering.)
3. I don't prohibit differential pricing for different services and classes of service.
4. I don't set any restrictions on contractual arrangements (apart from these two restrictions), including interconnection agreements or who pays. I think that should be left to private negotiation and competition.
5. I don't extend these requirements to other types of Internet providers such as backbone providers or those providing business services, as those are areas with plenty of competition.
6. I don't extend these requirements to wireless providers, because I think that with sensible market-based allocation of spectrum, there could be plenty of independent competition with much less capital expenditure than for wireline deployment.

I could possibly be persuaded that there is a place for common carriage requirements, especially for access circuits to businesses, which is where the last-mile providers could really engage in anti-competitive behavior against backbone providers that don't own a lot of last-mile wires (e.g., Level 3, Global Crossing, Sprint), now that the major telco last-mile providers have each merged with a major backbone provider themselves (Qwest/U.S. West, AT&T/SBC/BellSouth, Verizon/MCI). This requirement currently exists in the law for telcos, and unlike the common carriage requirement for DSL, is not planned to go away next year.

I would not put the above into the purview of the FCC, at least not with their current dispute resolution procedures which favor the telcos. Paul Kouroupas at Global Crossing (also my employer) has been arguing for "baseball-style" or final arbitration dispute resolution, where each side submits their best and final offer to an arbitrator, who chooses the best. This provides incentive for each side to try to reach the best agreement up front, as well as a process that can proceed quickly, without any government involvement or expense. This suggestion is the second point of Global Crossing's proposed REFORM legislative agenda. (Unbundling and common carriage of bottlenecks such as last-mile access circuits are the sixth point.)

Comments, criticisms? I should add that I believe what I've spelled out above is pretty close to what I've heard is in Sen. Stevens' telecom reform bill, though I haven't read it and I suspect he applies the nondiscrimination and unbundling requirements more widely than to residential broadband.

7 comments:

Lippard said...

Tom: Good question. Given the way I've written it, a provider could offer a plan such as you describe (pay for 10 Mbps, get an extra 5 mbps dedicated to MSN services)--but you would also still be able to pay for 15 Mbps of full Internet. You'd have to pay for that extra 5 Mbps in that case, though.

Thinking about it overnight, I'm not sure I wrote out my nondiscrimination provision with enough appropriate qualifications to permit the kind of anti-spam activity that I think providers need to be able to perform. Specifically, it needs to permit providers to ban spam-support services which may not directly cause disruption or degradation, but which are part of the supply chain for spam services. I would include in that providing DNS for spammers, web pages for spammed-for content, and distribution and sale of spamming tools the use of which is intended to violate the spam provisions of the AUP. (See the Global Crossing acceptable use policy for what I think a good AUP looks like.)

Further, I think a consumer ISP should be able to set restrictions on content hosted by their users which are more restrictive than just what's illegal, if they choose--this has no real anti-free speech effect since users will typically host web content at a web provider.

Lippard said...

I don't understand the concern about the Shaw offering. They are offering to allow you to get better-than-best effort service for third-party VoIP service for a fee. If you want to continue to use a third-party VoIP service on best-effort Internet, there's no charge--their offering in no way impacts the service that was previously available, but the offering gives the customer a way to ensure that their voice traffic packets will be the last to be dropped in circumstances of congestion within Shaw's network, which, given a lot of cable modem customers, is bound to occur from time to time. There are literally millions of compromised systems churning out spam, engaging in scan and sploit activity, and launching denial of service attacks, and better than 99% of them are sitting in people's homes, on consumer broadband connections.

What possible reason could there be to ban Shaw's offering (which strikes me as insane) or prevent them from charging for it? If you force them to allow anybody to put whatever traffic they want at a higher class of service, with no additional charge, then you've broken the QoS offering--everybody will just put all of their traffic at that level, rather than limiting it to traffic that actually has *need* to be there due to latency/jittery requirements.

There's a reason that toll-roads work, and congestion pricing works in downtown London. It causes those who don't really have the time constraints to defer their use to other times or places.

As for the bundled service offering, of course it can be offered cheaper--there's going to be a subsidy to the Internet component from the non-Internet component, just like advertising-subsidized content. If you want the advertising-free service, you're going to have to pay a premium.

As for the example in the link about the Internet component being ratcheted down to 1% of the bandwidth--I think that example is unrealistic because it won't offer consumers what they demonstrably want (see AOL), there's sufficient competition to prevent it, and if it were possible for providers to collude into pushing such a thing as the only option, it's quite obvious there would be a huge outcry, petitions to Congress, and lawsuits over it. No provider would want to deal with that.

I'm quite confident that best-effort Internet services are not going to become slower than today, even if they end up becoming a smaller percentage of the total pipe, unless actually interest in Internet declines. Given the history of communications, I think interest in services between individuals will continue to increase, not decrease, and providers who fail to cater to that will lose customers.

Lippard said...

Clarification: It's not Shaw's offering that strikes me as insane, it's the idea of prohibiting it.

I'll also note that my impression of Shaw's offering is that it is engineered the same way Global Crossng engineers QoS on its network for our VoIP traffic. See my posting on Dave Siegel and QoS in my net neutrality index for more detail on that.

Lippard said...

Mitch:

AOL didn't own the pipes, but neither did its competitors who drove it to open up full Internet access, the ISPs and Internet backbones. (BTW, AOL did own its own backbone, ANS. It traded ANS to Worldcom in exchange for the service and subscribers of CompuServe, and ended up rebuilding its own IP backbone, the AOL Transit Data Network, ATDN.) The last-mile telcos were pretty late to the game here, and have picked up the bulk of their Internet assets through acquisition and developing DSL, where they're still playing catch-up to cable (though they are catching up). Cable has a firm hold on higher-income customers, probably because they were the early adopters.

I think talk of "duopoly" exaggerates cooperation between cable and telco (they are fiercely competitive) and overlooks the growing role of wireless to create further competitive pressure.

Lippard said...

Mitch and Tom:

I think we already have some approximations to that model today.
For example, there are content distributors like Akamai and Limelight who have relationships with the eyeball customer providers, as well as content providers who already have private peering arrangements with the eyeball customer providers. Those who use Akamai or have private peering with the telcos and cable companies do already have an advantage above those who do not, and Akamai and Limelight provide a method for small players to get the equivalent of a lot more interconnections at a lower cost than negotiating all of those interconnections themselves.

I think that we will see this model extend to VPN-like services through inter-provider VPNs and inter-provider QoS. There will be mutual benefit in large providers doing settlement-free peering for higher classes of service, which will also provide competition for content providers to choose which large providers to connect to.

Mitch: The eyeball customers get to choose what to look at, but the content providers' spending (on number and location of server resources, and quantity and diversity of bandwidth) determines how many of them can look at one time. That's always going to be the case and will always give those with more money and resources the ability to have greater reach. The trick for the little guy has always been to find ways to piggy back on the bigger guys or work together with other little guys. This has always been true of every medium, from Internet to television to the printed page to the spoken word.

Lippard said...

Mitch, re your 2:06 p.m. post:

Do you see a possibility or likelihood of institution of a walled garden in the consumer broadband space? How would you see that happening? It seems to me that the moment one guy does it, they're going to severely alienate a huge percentage of their customer base and drive them to the competition, unless the competition does it at the same time. And if they both do it at the same time, the customers of both will raise holy hell through the courts.

Personally, given the choice between a 15 Mbps walled garden cable connection for free and a $200/mo business T1 with full Internet (that was the approximate local loop cost to my house about six or seven years ago), I'd take the latter.

Lippard said...

Rob: "Cox enters into a franchise agreement with the local municipality, but they use my property, not the city's, and I am not a party to those negotiations and have no right to refuse (nor am I compensated)"

This should not be the case. The city should only be able to grant CATV access to *public* rights of way and easements, not private property. If access to your private property is granted, then this should constitute a 5th amendment "taking" and be compensated as per Loretto v. Teleprompter Manhattan CATV Corp.

I agree that eminent domain should be restricted to genuine public use--I think that removing special privileges granted to telcos is desirable, as well as agree with you that it provides justification for treating them differently (i.e., regulate them) than if they came to hold those assets without government assistance.